Echinoid Community Structure and Rates of Herbivory and Bioerosion on Exposed and Sheltered Reefs

Home , African coral reefs, Bioerosion, Diadema savignyi, Echinothrix diadema

Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17

Contents lists available at ScienceDirect

Journal of Experimental Marine Biology and Ecology

journal homepage: www.elsevier.com/locate/jembe

Echinoid community structure and rates of herbivory and bioerosion on exposed and sheltered reefs

Omri Bronstein ⁎,YossiLoya

Department of Zoology, The George S. Wise Faculty of Life Sciences, Tel-Aviv University, Tel-Aviv 69978, Israel article info abstract

Article history: Echinoid–habitat relations are complex and bi-directional. Echinoid community structure is affected by the hab- Received 19 September 2013 itat structural and environmental conditions; while at the same time, echinoids may also act as ‘reef engineers’, Received in revised form 24 January 2014 able to alter marine environments on a wide geographic scale. In particular, echinoids play a major role in Accepted 9 March 2014 bioerosion and herbivory on coral reefs. Through feeding, echinoids reduce algal cover, enabling settlement Available online xxxx and coral growth. However, at the same time, they also remove large parts of the reef hard substrata, gradually Keywords: leading to reef degradation. Here, we compared coral and macroalgal abundance, echinoid community structure ﬁ Bioerosion and species-speci c rates of echinoid herbivory and bioerosion on reefs subjected to different intensities of oce- Coral reefs anic exposure. Spatio-temporal variations in coral and macroalgal cover were monitored, and populations of the Herbivory four most abundant echinoid species on the coral reefs of Zanzibar – Diadema setosum (Leske), Diadema savignyi MPAs (Michelin), Echinometra mathaei (de Blainville) and Echinothrix diadema (Linnaeus) – were compared between Sea urchins the Island's eastern exposed reefs and western sheltered ones. To account for the effect of management in the Western Indian Ocean context of reef exposure, we included marine protected areas (MPAs) of both types of reef categories (i.e. sheltered and exposed) in our comparison. Coral and macroalgal cover presented a conspicuous contrasting pattern across exposed and sheltered sites. While coral dominance and lack of macroalgae were prominent on sheltered reefs, an opposite trend of low coral cover and moderate–high macroalgal cover were found on exposed reefs.

Bioerosion was also significantly higher on exposed reefs than on sheltered ones (4.2–13 and 1.2–3.9 kg CaCO3 −2 −1 −2 −1 m year , respectively). The highest rates, recorded on Pongwe, with almost 7 kg CaCO3 m year , are among the highest echinoid bioerosion rates known to date. Management had a substantial effect on habitat and echinoid community structure, as coral cover was significantly higher, macroalgal cover lower, and echinoid densities generally reduced on MPAs regardless of exposure intensity. Our findings suggest that exposed reefs are susceptible to markedly higher degrees of echinoid bioerosion; however, adequate management measures can significantly reduce these rates, consequently altering the reef's trajectory for degradation. © 2014 Elsevier B.V. All rights reserved.

1. Introduction 1990; Hawkins and Lewis, 1982). At moderate sea urchin densities this action may facilitate a topographic complexity that favors increased Common coral-reef associated echinoids have a range of different biodiversity (Johnson et al., 2003) and may also enhance coral recruit- feeding modes. Echinoids are considered to be generalist herbivores as ment (Birkeland and Randall, 1981; Carpenter and Edmunds, 2006; their diets may include algae and seaweed (Klumpp et al., 1993; Griffin et al., 2003). However, at high sea urchin densities, echinoids Lawrence, 1975; Vaïtilingon et al., 2003), or omnivores due to the inclu- may limit reef growth through predation of coral tissue (Glynn et al., sion of animal tissue (Briscoe and Sebens, 1988; McClintock et al., 1982), 1979)orextensivecoral(Bak et al., 1984; Mokady et al., 1996) and crus- and even the occasional predation of live coral tissue (Bak and van Eys, tose coralline algae (CCA) erosion (O'leary and McClanahan, 2010). 1975; Carpenter, 1981; Glynn et al., 1979). This dietary flexibility, Moreover, the indiscriminate nature of echinoid grazing has a profound coupled with their great abundance on some coral reefs (Bauer, 1980; effect on coral community composition through its control of newly- McClanahan and Kurtis, 1991), place echinoids as keystone species in settled coral spat (Sammarco, 1980, 1982). Consequently, high sea ur- coral reef environments. As hard-substrate eroders (Bak, 1990; Glynn chin abundance may alter the structure of coral reef communities by et al., 1979; Hunter, 1977; Trudgill et al., 1987) they scrape the surface eroding the reef's coral framework, leading to gradual reef degradation. while grazing (Lawrence and Sammarco, 1982), reducing algal cover Many variables have been recognized as important in regulating (Mapstone et al., 2007) and breaking down reef substratum (Bak, echinoid food consumption. For example, species composition, body size, population densities (Bak, 1990, 1994; Carreiro-Silva and ⁎ Corresponding author. Tel.: +972 3 6409809; fax: +972 3 6727746. McClanahan, 2001; Scoffin et al., 1980), attraction to food (Vadas and E-mail address: [email protected] (O. Bronstein). Elner, 2003), hydrodynamics (Siddon and Witman, 2003), light (Mills

http://dx.doi.org/10.1016/j.jembe.2014.03.003 0022-0981/© 2014 Elsevier B.V. All rights reserved. O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 9 et al., 2000; Vaïtilingon et al., 2003), temperature (Larson et al., 1980), detailed account on echinoid community structure and associated rates and reproductive stage (Klinger et al., 1997), have all been mentioned of herbivory and bioerosion around the Island of Zanzibar, WIO. as factors influencing echinoid feeding rates and ecological impact. How- ever, beyond the physiological aspects determined by the life histories of 2. Methods particular species, echinoid food consumption, and consequently the rates of herbivory and bioerosion, must be considered in terms of the en- 2.1. Study sites vironmental conditions that exist in their habitats, as gradients in the physical environment may produce variability in the abundance and dis- Coral communities and associated echinoid populations were tribution of echinoid populations (Andrew, 1993; Clemente and studied on six reefs surrounding Zanzibar Island (Fig. 1). The sites Hernández, 2008). Several studies have investigated the relationship be- were selected to represent sheltered and exposed reefs in terms of oce- tween coral reef associated echinoids and their habitat (e.g., Dumas et al., anic exposure. To test for effects of marine protected areas, MPAs from 2007; Graham and Nash, 2013; McClanahan, 1998; McClanahan and both exposure categories (i.e. sheltered and exposed) were selected. Kurtis, 1991; O'leary and McClanahan, 2010; Peyrot-Clausade et al., However due to the scarcity of MPAs in the region, only one such site 2000). These publications suggest aspects such as structural complexity, per exposure category was available for this analysis. Three sites, macroalgal and coral cover, sedimentation, and the presence or absence Bawe (06°08.7′S; 039°08.2′E), Changu (06°06.8′S; 039°09.8′E), and of predators, as having substantial effects on the composition, distribu- Chumbe (06°16.3′S; 039°10.2′E), were selected on the sheltered tion, and size of related echinoid populations. For example, marine western side of the main island facing the Zanzibar channel. The site protected areas (MPAs) protecting various echinoid predators conse- at Changu is located ca. 5.5 km from Zanzibar Town and a similar quently present lower rates of sea urchins compared to reefs with distance from the site at Bawe. Chumbe is located ca. 12 km south depauperate predatory populations (McClanahan and Kurtis, 1991; of Zanzibar Town, and has been a private nature reserve, developed McClanahan et al., 1999). Additionally, echinoid communities tend to and managed by the Chumbe Island Coral Park (CHICOP), since display strong differences in species distribution between exposed and 1992 (Nordlund and Walther, 2010). The sites on the exposed eastern sheltered reefs, making sea urchin ecology further complex (Dumas side of Zanzibar were Kiwengwa (06°00.9′S; 039°24.6′E), Pongwe et al., 2007). (06°01.9′S; 039°25.2′E), and Mnemba (05°48.5′S; 039°21.3′E). The Zanzibar Island (Unguja, Tanzania) is situated on the continental shelf of Tanzania between 50°40′ and 60°30′ south of the equator, 35 km from the mainland. Being an island surrounded by coral reefs, ex- N posed to strong easterly winds and with a sheltered west coast, makes Zanzibar an ideal study location for echinoid ecology. Located off the AFRICA East-African shoreline, the island's coral reefs are fundamental to the entire marine environment and of great economic importance for the Mnemba large human population that depends on them for a livelihood (Jiddawi, 1997; Khatib, 1997; Mbije et al., 2002; Ngoile and Horrill, 1993). Small patches of mangrove forest and shallow patches of fringing reefs occur along the more sheltered western coast, while on the more exposed eastern coast fringing reefs slope up to a narrow coastal lagoon backed by sand beaches or fossil coral cliffs (Richmond, 2002). The eastern and western sides of the island are subject to markedly different wave and current intensities; reefs on the eastern ocean-facing side are exposed to the Indian Ocean (IO) and are susceptible to strong waves and currents, while reefs in the Zanzibar channel, on the Island's Kiwengwa western side, are sheltered from direct exposure to the IO (Bergman and Öhman, 2001; Ngoile, 1990). Swell waves generated in the IO can travel Pongwe undisturbed for thousands of miles before hitting the Island's eastern ZANZIBAR reefs. These swell waves occur off the east coast of Zanzibar for much of the year, changing their orientation from north-east (between Octo- Bawe ber and March) to south-east (between March and October) depending Changu on monsoonal season (McClanahan, 1988b; Zanzibar Department of Environment and MACEMP, 2009). In contrast to the north-east monsoon, the south-east monsoon is characterized by high cloud cover, rain, high wind energy, decreased temperatures and light, and rougher seas, with velocities of the East African Coastal Current (EACC) increasing to a speed of four knots (McClanahan, 1988b). The semi-diurnal tides have mean spring amplitude of 3.3 m, with associated tidal cur- Chumbe rents being stronger on the east coast, where currents up to three knots are common (Bergman and Öhman, 2001). Here, we studied coral and macroalgal cover, echinoid community structure and associated rates of herbivory and bioerosion on exposed and sheltered coral reefs. The following hypotheses were tested: (1) Coral and macroalgal cover vary between exposed and sheltered reefs. (2) Echinoid community structure, and consequently the intensity of echinoid-induced bioerosion, are influenced by the degree of oceanic 10 KM exposure (e.g., the exposure to higher intensities of waves, currents, tides, etc.). (3) Rates of echinoid herbivory and bioerosion on marine- protected areas are lower than on unprotected sites. Finally, we present Fig. 1. Map of Zanzibar showing the six study sites. Double circles indicate sites are data on spatio-temporal variations of coral and macroalgal cover, and a marine-protected areas. 10 O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 former two sites are located ca. 3 km from one another on the reef-flats burned at 550 °C for 6 h in order to eliminate the organic material. of Zanzibar's eastern fringing reefs. Mnemba is located ca. 2.5 km off After drying, the samples were weighed again. The difference in weights Zanzibar's north-eastern tip, and has been protected from extractive re- before and after ashing was used as a measure of the organic matter in source use since 1989 (Bergman and Öhman, 2001). the sea urchin's gut. The CaCO3 fraction was then determined by acidic digestion of the residual material after removal of the organic matter.

2.2. Coral and macroalgal cover 1.13 N hydrochloric acid (HCl) was used to dissolve the CaCO3.0.5g subsamples of residual ashed material were incubated for 10 min with Three surveys were conducted between 2006 and 2008 (November 25 ml 1.13 N HCl. Following incubation and complete dissolution of

2006, March 2007, and April 2008). At each site, 12 20 × 1 m belt- CaCO3, the solution of HCl and residual material was filtered on transects were randomly placed to sample the composition of coral as- 0.22 μm PTFE filters (Millipore Hydrophilic Durapore Membrane disk semblages at depths of 3–5 m using SCUBA. Each belt-transect was filters), using a suction filter system. Prior to filtration the filters were subdivided into 20 1 × 1 m quadrats, which were photographed with dried in a pre-heated oven at 60 °C for 24 h, and weighed to obtain a digital camera (Olympus C-5060) attached to a 1 × 1 m PVC-frame. the filter's dry weight. Following filtration the filters were dried as be- The percentage coral and macroalgal cover was calculated from the fore and reweighed. The weight of the residual material retained in photographs taken in the field. the filter was calculated by subtracting the filter's dry weight before filtration from the dry weight after filtration. This weight corresponds to 2.3. Echinoid community structure the non-soluble residue fraction. Subtracting the weight of the non- soluble residue fraction from the total ashed weight resulted in the

Species assemblages and densities were quantified in March 2008. weight of CaCO3 in the gut. The results obtained in the analysis are Ten randomly placed belt-transects, measuring 20 × 0.5 m, were con- expressed in terms of percentage of each fraction in the sea urchin gut ducted at each site, at depths of 1–5 m using SCUBA. All echinoids with- for each species at each site. in a transect were recorded and identified to the species level. When present, the size frequency distributions (SFD) of the dominant species 2.5. Estimating the rates of bioerosion and herbivory (i.e. Diadema setosum, Diadema savignyi, Echinometra mathaei and Echinothrix diadema) from each site were estimated. Estimations were The daily food consumption could be estimated based on two based on length measurements of the first 100 individuals encountered parameters: the average amount of food in the gut, and the number of in randomly placed one square meter quadrates. The length was mea- hours necessary for complete gastric evacuation (Bajkov, 1935; Elliott, sured underwater to the nearest 0.5 mm as the longest axis at the 1972; Elliott and Persson, 1978). The average amount of food in the ambitus, using thin blade Vernier calipers. The mean size of individuals gut was calculated for each species at each site as previously described. in the population was then calculated for each species at each site. The rates of gastric evacuation for the echinoid species studied were obtained from Carreiro-Silva and McClanahan (2001). To calculate the 2.4. Echinoid gut contents analysis daily ingestion rates, the average amount of food in the gut and the rates of gastric evacuation were used in an equation developed by Individual rates of bioerosion and herbivory of the dominant sea ur- Elliot and Persson (1978). Assuming an exponential rate of gastric evac- chin species at each site were determined in experiments conducted at uation and a constant rate of food consumption, the daily rates of food the Institute of Marine Sciences (IMS) in Zanzibar City during March consumption could be calculated using the equation: 2008. Throughout this study, the term ‘bioerosion’ refers to the total amount of newly eroded CaCO3 from the hard reef substrate, which is F ¼ CR largely composed of scleractinian corals and crustose coralline algae (CCA). These rates, together with species densities, were then used to where the daily food consumption (F) could be estimated from the av- estimate annual echinoid bioerosion and herbivory rates per square erage amount of food (C) in the stomach at the time of sampling, and the meter for each species at each site. The total bioerosion and herbivory rate of gastric evacuation (R) (see Carreiro-Silva and McClanahan, rates for the entire Zanzibar region were evaluated by pooling all sites 2001). together. The two assumptions at the basis of this model – an exponential rate Ten individuals of each species from each site were collected from of gastric evacuation and a constant rate of food consumption – were pre- the field and brought to the lab. Sampled individuals were chosen ac- viously validated for the echinoid species of the current study. An expo- cording to the mean sizes of each species at the different study sites, nential rate of gastric evacuation was demonstrated in all species of the as determined by the size frequency distribution (SFD) estimations current study by conducting gut emptying experiments (Carreiro-Silva noted above. Collection was carried out on the reef-flats, in early morn- and McClanahan, 2001; McClanahan and Kurtis, 1991; Mokady et al., ing, at a depth of 3 m, using SCUBA diving. Individual sea urchins were 1996). The assumption of a constant rate of food consumption is widely separately placed in sealed containers underwater to avoid loss of mate- accepted (e.g. McClanahan and Kurtis, 1991; Mokady et al., 1996), and rial during transfer. At the lab, the sea urchins were measured and is supported by field observations (Carreiro-Silva and McClanahan, weighed after being blotted for five minutes on filter paper. They were 2001; Glynn et al., 1979; Klumpp et al., 1993)aswellascontrolledfield then dissected under a binocular to extract the gut wall from the gut experiments (Downing and El-Zahr, 1987), and is in agreement with contents. Extractions were followed by repeated rinses with distilled our own observations of sea urchins actively feeding during all hours of water. The gut contents were then analyzed in terms of organic and in- the day. organic fractions, with the latter being further separated into calcium To estimate the true scale of reef degradation (i.e. the scraping off of carbonate (CaCO3) and non-soluble residue fractions (e.g. quartz grains, new material from the reef's hard substratum), the source of CaCO3 sponge spicules, and silt). Analysis of the organic fraction followed a found in the sea urchins' guts must be considered (Scoffin et al., modification of the ignition-loss method (Dean, 1974). The total ex- 1980). It is therefore essential to distinguish between reworked sedi- tracted gut content was dried in a preheated oven (WTB Binder 1505) ment (i.e. recycling of previously eroded sediment) and newly-eroded at 60 °C for 48 h (or until a stable weight was reached) and weighed sediment (Bak, 1990; Hunter, 1977; Mokady et al., 1996). The measure with a Shimadzu AW220 analytical balance to the nearest 0.0001 g. of reworked sediment can then be subtracted from the total amount of

The low drying temperature of 60 °C was necessary to minimize the CaCO3 in the gut, and the remaining portion considered as a direct rep- loss of volatile, especially lipoid, constituents. The samples were then resentation of the reef's framework erosion. However, to date there is transferred to a furnace (Carbolite 1200 °C Ashing-plus furnace) and little consensus over the way to adequately estimate the amounts of O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 11 echinoid sediment consumption (i.e. reworked sediment). Mainly, it performed. Permutations were performed using the lmPerm package has not been concluded whether sediment ingestion by echinoids is a (Wheeler, 2010) for data analysis, allowing all permutations of Y (i.e. process driven by behavior, or merely an outcome of environmental Perm = “Exact”). Temporal variations in coral and macroalgal cover conditions (e.g., sediment load). Our estimates of the proportion of were tested using permutation analysis of variance (pANOVA), and spa- reworked sediment and gut turnover rates for D. setosum, D. savignyi, tial differences (i.e. between sites and sea facing sides) using a nested and E. diadema are based on Carreiro-Silva and McClanahan (2001), design pANOVA with year b site b side. Differences in sea urchin densi- and for E. mathaei from McClanahan and Kurtis (1991). These estimates ties were tested using a two-way pANOVA with sites and species as fac- represent the highest values of reworked sediment for these species tors. Size frequency distributions were compared using pair-wise available from the literature and as such would yield the most conserva- Kolmogorov–Smirnov tests and adjusted for multiple resting using the tive (i.e. low) bioerosion estimations. Still, as the impact of sedimenta- Bonferroni correction to minimize false-discovery-rate. Gut content tion on echinoid sediment consumption could not be elucidated at fractionation was tested using one-way pANOVA. The Tukey Honest Sig- this point, these values should be treated with caution. niﬁcant Difference (HSD) method which controls for the Type I error rate across multiple comparisons was used when appropriate. 2.6. Statistical analysis 3. Results Data analyses were performed using R software for statistical computing (Team, 2013). All data were tested for normality and homogene- 3.1. Coral community structure ity of variance prior to deciding upon the appropriate statistical test. As data violated test assumptions of normal distribution and homoscedas- Western sites presented signiﬁcantly higher coral cover in compari- ticity, and as data transformations failed to bring the data to meet the son to eastern sites (pANOVA, p b 0.01; Fig. 2A). Throughout the years assumptions of parametric statistical tests, permutation analysis was of the surveys trends of coral cover have remained consistent within

A Coral cover

80 a Nov- 06 W a E 70 a Mar- 07 Apr- 08 60 a a 50 a a a 40 a

20 a ab % coral cover (AVG ± SE) 10 b a a a a a a 0 Changu Bawe Chumbe Mnemba Kiwengwa Pongwe Site

B Macroalgae cover

70 a W E a Nov- 06 60 Mar- 07 Apr- 08 50 a c 40

20 b b 10 % macroalgae cover (AVG ± SE) a a a a a a a a a a ab a 0 Changu Bawe Chumbe Mnemba Kiwengwa Pongwe Site

Fig. 2. Coral (A) and macroalgal (B) percent cover (mean ± SE) at the different study sites between November 2006 and March 2008. Western sites and eastern sites are denoted W and E, respectively. Lowercase letters above bars indicate per-site significance groupings as inferred from Tukey HSD analyses. 12 O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 sites (pANOVA, p N 0.05 for all sites) except for the site of Mnemba Analyses of size frequency distributions revealed significant differ- where a constant decline in coral cover had been recorded (pANOVA, ences in size distributions (Table S1 in Electronic Supplementary p b 0.01), displaying the sharpest reduction of coral cover between Material; ESM) and average sea urchin sizes among species (Table 1) March 2007 and April 2008. Of the western sites, Chumbe recorded and sites (Table 2, Fig. S1). Sea urchins from the eastern sites were larger the highest coral cover (64%–72%) throughout the three years of the than their western conspecifics. For example, the mean diameter of survey. The lowest coral cover was recorded on Kiwengwa, with cover E. diadema was highest on Pongwe, followed by Changu and Bawe less than 0.5%. Macroalgal cover presented an opposite trend to coral (ca. 100, 96, and 78 mm, respectively) (pANOVA, p b 0.001). Similarly, cover (Fig. 2B). While the western sites presented negligible macroalgal E. mathaei presented relatively small individuals on Changu and Bawe cover, the eastern sites presented significantly higher algal cover (ca. 36 and 34 mm, respectively), while those from Kiwengwa and (pANOVA, p b 0.001), reaching more than 50% at some locations Pongwe were significantly larger (47 and 38 mm, respectively) (Fig. 2B). The highest macroalgal cover throughout the duration of the (pANOVA, p b 0.001). D. savignyi was significantly larger on Bawe surveys (ca. 35%–52%) was recorded on Kiwengwa on the Island's east- compared to Changu and Pongwe (pANOVA, p b 0.001). D. setosum pre- ern exposed side. Temporal variations in macroalgal cover indicate no sented significantly smaller individuals than their conspecifics on change at the western sites (pANOVA, p N 0.05 for all sites), and signif- Changu and Bawe (pANOVA, p b 0.002). Sea urchin densities on the icant increases in percentage cover for both Mnemba and Kiwengwa eastern MPA at Mnemba were too low to adequately perform size fre- (pANOVA, p b 0.001 for all; Fig. 2B). In the Mnemba MPA macroalgal quency estimations. cover increased seven-fold from March 2007 to March 2008, and at Pongwe in macroalgal cover constantly increased from ca. 1% to 9% and 33% from 2006 to 2008. 3.3. Echinoid gut contents analysis

3.2. Echinoid community structure In all four species, the inorganic portion was larger than the organic

portion (Fig. 4). Interspecific differences were found in the CaCO3 and Sea urchin populations varied in species assemblages and popula- organic matter portions, where E. diadema had a significantly larger pro- tion densities both within and between sites, and across the western portion of organic matter and a smaller proportion of CaCO3 than all and eastern sides of the Island (Fig. 3). Variations in densities were sig- other species (pANOVA, p b 0.01). D. setosum presented significantly nificant between sites (pANOVA, p b 0.001), and species (pANOVA, higher CaCO3 content and lower organic matter than all other species p b 0.001). D. setosum dominated the western sheltered sites (i.e. (pANOVA, p b 0.001). No significant differences in the non-soluble por- Changu, Bawe and Chumbe) followed by E. mathaei which instead dom- tions were observed among species (pANOVA, p N 0.05). inated the exposed eastern ones (i.e. Kiwengwa and Pongwe) (Fig. 3). Intraspecific comparisons of gut contents revealed significant differ-

E. mathaei densities were more than seven-fold higher on the eastern ences in the proportions of CaCO3 and organic matter among species be- sites than on the western ones (ca. 14 and 2 ind m−2,respectively), tween sites (Fig. 4). However, while species like E. mathaei presented but were always absent from marine-protected areas on both sides. increased proportions of organic matter on exposed reefs in comparison The two MPAs, Mnemba (east), and Chumbe (west), differed from to sheltered reefs, the results for the other species were not so clear. The neighboring sites in both species assemblages and densities (Fig. 3). proportion of organic gut content of E. diadema on Changu was about For the eastern reefs Mnemba presented low sea urchin densities 40%, compared to less than 20% on adjacent Bawe. The organic content (1.62 ± 1.0 ind m−2) and no E. mathaei, a sharp contrast to the in the gut of E. diadema on the only exposed reef with sufﬁcient sea ur- high echinoid densities at Kiwengwa and Pongwe (20.50 ± 12.0 and chin abundances, Pongwe, was around 20%. D. savignyi displayed a re- 30.19 ± 10.6 ind m−2, respectively) (pANOVA, p b 0.001; Fig. 2). verse pattern with Bawe around 25% organic gut composition On the western side, Chumbe had low density of E. mathaei compared compared to 10% on Changu and 15% on Pongwe. D. setosum presented to Changu and Bawe (pANOVA, p b 0.05) but similar densities of similar CaCO3 and organic matter proportions on all sheltered sites D. setosum (pANOVA, p N 0.05). (~10%) (Fig. 4).

40 W E A 35 a AB 30 a

(AVG ± SE) 25 -2

20 BC

15 ab CD CD

10 a a a D b b

# of individuals m 5 b a b b b b b b b b bbb b b b b b bbb b 0 Changu Bawe Chumbe Mnemba Kiwengwa Pongwe Sites

Diadema setosum Diadema savignyi Echinometra mathaei Echinothrix diadema Other

Fig. 3. Sea urchin densities at the different study sites around the Island of Zanzibar. Densities were measured in 20 m × 0.5 m belt transects (n = 10 transects per site). Bars indicate average species speciﬁcdensitiesperm−2 (mean ± SD) from surveys conducted in March 2008. Western sites and eastern sites are denoted W and E, respectively. Signiﬁcance groupings as inferred from Tukey HSD analyses are presented as lowercase letters to indicate groupings among species within sites, and as uppercase letters among sites. O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 13

Table 1

Test diameters, gut dry weights, daily rates of food ingestion, herbivory and CaCO3 ingestion rates, and bioerosion rates for the mean-sized echinoid (mean ± SE) species around Zanzibar. Diadema setosum (n =30),Diadema savignyi (n =30),Echinothrix diadema (n = 30) and Echinometra mathaei (n = 40). *Data obtained from Carreiro-Silva and McClanahan, 2001.

Species Test diameter Gut weight Gut turnover Percentage of Total ingestion Herbivory rate CaCO3 ingestion rate Bioerosion rate −1 −1 −1 −1 −1 −1 −1 (mm) (g) rate (day )* reworked sediment rate (gut dry weight (g algae ind day ) (g CaCO3 ind day ) (g CaCO3 ind day ) in the urchins gut* ind−1 day−1)

D. setosum 45.27 ± 0.34 3.42 ± 2.20 1.18 73.5 4.03 ± 0.41 0.40 ± 0.07 3.59 ± 0.37 0.95 ± 0.10 D. savignyi 53.76 ± 0.32 2.84 ± 1.10 0.89 72.0 2.53 ± 0.29 0.50 ± 0.13 2.00 ± 0.18 0.56 ± 0.05 E. diadema 91.55 ± 1.97 13.53 ± 2.30 1.14 69.7 15.42 ± 2.25 3.87 ± 0.78 11.43 ± 1.59 3.46 ± 0.48 E. mathaei 38.81 ± 0.04 0.36 ± 1.65 1.72 0 0.62 ± 0.06 0.13 ± 0.02 0.48 ± 0.05 0.48 ± 0.05

3.4. Rates of ingestion, herbivory, and bioerosion fundamental variables: species identity, body size and population densities (Bak, 1994). These variables are both governed by the habitat's phys- Significant differences in rates of ingestion, herbivory and bioerosion ical conditions and at the same time, influence the structure and were found between sites and species. The greatest rates of ingestion composition of the habitat itself (reviewed by Steneck, 2013). For exam- were found for E. diadema, with a daily CaCO3 consumption rate ple, factors such as habitat structural complexity (Graham and Nash, −1 −1 of (11.43 ± 1.6 g CaCO3 ind day ), more than three-fold that 2013), exposure to surf (Ebert, 1982), and regulation by predation −1 −1 of D. setosum (3.59 ± 0.4 g CaCO3 ind day ), 5.7-fold that of (McClanahan and Kurtis, 1991; McClanahan and Shafir, 1990), are all −1 −1 D. savignyi (2.00 ± 0.2 g CaCO3 ind day ), and almost 24-fold that known to influence echinoid community structure on coral reefs. On −1 −1 of E. mathaei (0.48 ± 0.05 g CaCO3 ind day )(Table 1). E. diadema the other hand, reef degradation through bioerosion is often associated also had the greatest rates of herbivory per day (3.87 ± 0.8 g algae with high sea urchin abundance (Bak, 1990; Scoffin et al., 1980), direct ind−1 day−1) than any of the other species; 7.7-fold more than coral predation (Carpenter, 1981; Glynn et al., 1979), and echinoid D. savignyi (0.50 ± 0.1 g algae ind−1 day−1), 9.7-fold more than control of newly-settled spat (Sammarco, 1980, 1982). The type and in- D. setosum (0.40 ± 0.07 g algae ind−1 day−1), and almost 30-fold tensity of environmental conditions are likely to regulate echinoid com- more than E. mathaei (0.13 ± 0.02 g algae ind−1 day−1). E. diadema munities both directly (e.g., inability of certain species to resist strong also presented the highest individual rate of bioerosion (3.46 ± 0.5 g currents) (Tuya et al., 2007), and indirectly through shaping coral reefs −1 −1 CaCO3 ind day ). D. setosum average bioerosion rate (0.95 ± 0.1 g structure to either favor or exclude specific echinoid species (Dumas −1 −1 CaCO3 ind day ) was less than a third of that of E. diadema while et al., 2007; McClanahan and Kurtis, 1991; McClanahan and Shafir, almost twice the rate of D. savignyi and E. mathaei (0.56 ± 0.05 and 1990). −1 −1 0.48 ± 0.05 g CaCO3 ind day , respectively) (Table 1). Gross annual bioerosion was higher on the eastern exposed reefs 4.1. Coral and macroalgal cover in comparison to the western sheltered ones (ANOVA, F = 52.97, p b 0.001; Fig. 5). The highest bioerosion rates were recorded on Pongwe, We found marked differences in coral and macroalgal cover, echi- −2 with more than 6.9 kg CaCO3 m eroded annually. Kiwengwa presented noid community structure, and rates of herbivory and bioerosion be- −2 −1 the second highest bioerosion levels (4.2 kg CaCO3 m year ), slightly tween exposed and sheltered reefs. That coral cover is low in areas of −2 −1 higher than Bawe (3.9 kg CaCO3 m year ) and twice the erosion rates high algal cover has been thoroughly discussed in the scientific litera- −2 −1 at Changu (2.1 kg CaCO3 m year ). The lowest bioerosion rates were ture, and is often associated with algal predominance over scleractinian −2 −1 recorded on Chumbe, with only 1.2 kg m year CaCO3 eroded corals in competition for environmental resources (reviewed by (Table 2, Fig. 5). McCook et al., 2001). Respectively, our data present opposite trends of coral and algal cover: where coral cover was low, algal cover was high 4. Discussion and vise versa. This pattern was evident in all of the sites studied regardless of the level of ocean exposure or protection (i.e. MPA and non- The magnitude of echinoid grazing, and consequently the rates of MPA), reflecting the generality of these coral–algal interactions. None- herbivory and bioerosion, are believed to be determined by three theless, between-site differences corresponded to the degree of habitat

Table 2

Densities, test diameters, herbivory rates, CaCO3 ingestion rates, and bioerosion rates (mean ± SE) of the dominant echinoid species of Zanzibar. Data presented by site and are based on the average test diameters of each species in every site (n = 10 individuals per species in each site).

Site Sea facing/ Species Density Test diameter Herbivory rate CaCO3 ingestion Bioerosion rate Net herbivory Net bioerosion Total site −2 −1 category (ind m ) (mm) (g algae ind rate (g CaCO3 (g CaCO3 (kg algae (kg CaCO3 bioerosion −1 −1 1 −1 −1 −2 −1 −2 −1 day ) ind day ) ind day ) m year ) m year ) (kg CaCO3 m−2 year−1)

Changu West (sheltered) D. setosum 5.99 ± 0.56 46.45 ± 1.00 0.31 ± 0.04 2.97 ± 0.26 0.79 ± 0.07 0.68 ± 0.09 1.73 ± 0.15 2.10 ± 0.08 D. savignyi 0.44 ± 0.07 52.69 ± 0.53 0.15 ± 0.01 1.36 ± 0.11 0.38 ± 0.03 0.02 ± 0.00 0.06 ± 0.00 E. diadema 0.24 ± 0.01 96.19 ± 0.85 2.47 ± 0.72 3.75 ± 1.09 1.14 ± 0.33 0.22 ± 0.06 0.10 ± 0.03 E. mathaei 1.26 ± 0.36 35.65 ± 0.80 0.05 ± 0.01 0.46 ± 0.10 0.46 ± 0.10 0.02 ± 0.00 0.21 ± 0.05 Bawe West (sheltered) D. setosum 5.72 ± 0.87 47.52 ± 1.15 0.69 ± 0.17 5.82 ± 0.53 1.54 ± 0.14 1.44 ± 0.36 3.22 ± 0.29 3.90 ± 0.15 D. savignyi 0.4 ± 0.06 55.55 ± 0.55 1.07 ± 0.33 2.55 ± 0.39 0.72 ± 0.11 0.16 ± 0.05 0.11 ± 0.02 E. diadema 0.18 ± 0.12 78.31 ± 0.60 1.62 ± 0.55 8.38 ± 0.63 2.54 ± 0.19 0.11 ± 0.03 0.17 ± 0.01 E. mathaei 4.79 ± 2.80 34.03 ± 0.84 0.04 ± 0.00 0.23 ± 0.04 0.23 ± 0.04 0.07 ± 0.00 0.40 ± 0.07 Chumbe West (sheltered) D. setosum 6.38 ± 0.57 41.85 ± 1.36 0.20 ± 0.02 1.97 ± 0.26 0.52 ± 0.07 0.47 ± 0.04 1.21 ± 0.16 1.21 ± 0.16 Kiwengwa East (exposed) E. mathaei 20.28 ± 3.63 47.42 ± 0.86 0.24 ± 0.02 0.57 ± 0.10 0.57 ± 0.10 1.78 ± 0.14 4.22 ± 0.23 4.22 ± 0.23 Pongwe East (exposed) D. savignyi 1.65 ± 0.57 53.06 ± 0.68 0.29 ± 0.02 2.08 ± 0.30 0.58 ± 0.08 0.17 ± 0.02 0.35 ± 0.09 6.91 ± 0.63 E. diadema 1.41 ± 0.26 100.16 ± 0.99 7.52 ± 1.69 22.17 ± 1.56 6.72 ± 0.47 3.87 ± 1.69 3.46 ± 0.47 E. mathaei 12.47 ± 3.98 38.14 ± 1.10 0.17 ± 0.02 0.68 ± 0.12 0.68 ± 0.12 0.77 ± 0.16 3.10 ± 0.99 Mnemba East (exposed) NA NA NA NA NA NA NA NA NA 14 O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17

100 W E 80

60 6 40 Changu 20 ) (AVG ± SE)

0 −1

100 yr −2

80 m

3 4

40 Bawe

0 2 100

60 NA NA NA 40 Chumbe

20 Bioerosion rates (kg CaCO 0

0 Changu Bawe Chumbe Mnemba Kiwengwa Pongwe 100 Sites 80

60 NA NA NA Diadema setosum Diadema savignyi Echinometra mathaei Echinothrix diadema

% of gut contents 40

20 Kiwengwa Fig. 5. Echinoid annual bioerosion rates per site and species at six sites surrounding the −2 0 Island of Zanzibar. Bars indicate mean ± SE bioerosion rates per m of each species (color coded) and every site (stacked). Western sites and eastern sites are denoted 100 W and E, respectively. The number of samples collected was n = 10 per species per site. 80

60 NA

Pongwe of herbivorous ﬁsh may compensate for the lack of echinoids in regulat- 20 ing algal cover at that site. Thus, though the reason for Mnemba's coral 0 cover loss is not resolved at this stage, it might be a consequence of glob- 100 al changes and the overall large-scale coral decline (Bellwood et al., 80 2004; Hughes, 2003). In contrast coral cover on the Chumbe MPA was

60 exceptionally high throughout the duration of the study. Some of the mechanisms that work to maintain such high coral cover may be attrib- 40 ﬁ All sites uted to the extensive and ef cient protection these reefs receive (e.g., 20 no take zone, diving restrictions and strict anchoring regulations). How-

0 ever, in contrast to the reefs on Changu and Bawe, the geographical lo- E.mathaei E.diadema D. savignyi D. setosum cation of Chumbe, away from the big urban center of Zanzibar Town, is Species likely to further reduce indirect anthropogenic stressors, such as water pollution, that may also contribute to the prosperity of the latter reefs. Calcium carbonate Non- soluble Organic matter 4.2. Echinoid community structure Fig. 4. Fractionation of gut contents (mean ± SE) at each study site, and the pooled data per species of the four dominant echinoid species on the coral reefs of Zanzibar. Color in- Multiple environmental, ecological, and biological factors concur- dicates: CaCO (dark gray), non-soluble residues (light gray) and organic matter (black) 3 fi fractions. (n = 10 individuals per species, per site). rently occurring on coral reefs make it dif cult to elucidate the resulting echinoid species distributions (Dumas et al., 2007; Johansson et al., 2013). Nonetheless, echinoid community structure in terms of species composition, densities, and average body size, varied considerably be- exposure: while exposed reefs had high algal and low coral cover, tween sheltered and exposed reefs. Of the four dominant echinoid spe- sheltered reefs presented an opposite trend. It appears that coral com- cies found around the Island of Zanzibar, the two most abundant munities on Zanzibar's exposed reefs are being competitively excluded species, D. setosum, and E. mathaei, were also the most affected by the by algae, while on sheltered reefs coral communities are sustained level of habitat exposure. While D. setosum was prevalent on sheltered through algal regulation. Of particular interest is the Mnemba MPA on sites, it was absent on exposed sites where E. mathaei predominated the north-eastern exposed side. Our data show a significant decrease (Fig. 3). D. setosum is known to be restricted to quiet or protected waters in coral cover and, at the same time, a significant increase in macroalgal (Ebert, 1982). In exposed areas, the large overall volume (body and cover between 2007 and 2008 (Fig. 2). Though coral cover at this site is spines) of Diadema may prevent it from resisting the strong currents historically low (Bergman and Öhman, 2001; Ngoile, 1990), our mea- and surf, causing it to detach from the substrate, as demonstrated in surements reflect the lowest ever recorded coral cover in what is con- Diadema on the reefs of the Canarian Archipelago (Tuya et al., 2007). De- sidered Zanzibar's oldest established MPA (Obura et al., 2002). spite high macroalgal cover (Fig. 2B) and presumably low predation Nonetheless, management at this site seems effective, judging by the pressure as a result of intense fishing activity in the area (Jiddawi and abundance of herbivorous and predatory fish on the site (Brokovich, Öhman, 2002; Ngoile et al., 1988), D. setosum spatial distribution is pers. comm.). The abundance of predatory fish may account for the most likely a result of physical variables (e.g., surf, currents, etc.) rather lack of sea urchins (McClanahan and Kurtis, 1991), while the abundance than biotic factors (e.g., predation, food availability, etc.). O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 15

In contrast to D. setosum, E. mathaei, though present at most western filamentous algae on loose sediments (Bronstein, unpublished data). sites, predominated the eastern exposed sites (Fig. 3). Similar to Russo's Such activity would imply a nutritional strategy for utilizing an available findings from reefs with stronger water flow (Russo, 1977), our data resource through intentional sediment ingestion. These observations, show higher densities and a larger average body size of E. mathaei on despite being far from conclusive, should serve to underline gaps in the exposed reefs. These patterns of size and abundance might be attrib- our understanding of this process, and stress the need to further eluci- uted to food limitations and availability in the two different reef date the issue of echinoid reworked sediment evaluations. environments. As E. mathaei is generally sedentary, limited to the prox- Sheltered sites experienced lower degrees of bioerosion in compar- imity of its burrows (Langdon, 2012; Young and Bellwood, 2011), in ison to exposed sites. Of the sheltered sites, the highest degree of late successional algal communities it was suggested to rely on drift bioerosion was recorded on Bawe, being over three-fold higher than algae for a large part of its diet (Johansson et al., 2013; McClanahan and that recorded on Chumbe, and almost twice as that of Changu Muthiga, 2001; Russo, 1977). The significantly higher macroalgal cover (Table 2). Some of these differences may be attributed to differences and stronger currents on the eastern exposed reefs generate large in species composition and abundance on these reefs. While in Bawe amounts of drift algae, thus increasing food accessibility for populations four different echinoid species contributed to the total bioerosion, in of E. mathaei. However, in comparison to D. setosum, E. mathaei's distribu- Chumbe D. setosum was the sole contributor, and although similar spe- tion is thought to be less susceptible to environmental variables (Dumas cies compositions were observed on Bawe and Changu, the abundance et al., 2007). E. mathaei's proliferation throughout the exposed unprotect- of E. mathaei on Bawe was 3.8 times higher than on Changu. Gross ed sites might also be attributed to its ability to competitively exclude bioerosion on Bawe is still more than two-fold higher than that on other echinoid species (McClanahan, 1988a). In this context of inter- Chumbe, even when only D. setosum is considered, and despite similar specific competition, variations in the echinoid guild of the Bawe site abundance and algae cover on Chumbe. Apparently the higher gross will be interesting to follow during the next few years, as E. mathaei den- bioerosion on Bawe may instead be attributed to the higher average sities there are constantly rising (Bronstein, unpublished data) and are body size of D. setosum on these reefs, as rates of food consumption now as high as those of D. setosum (Fig. 3, Bronstein, unpubl. data). are known to increase with sea urchin size (Bak, 1990; Klumpp et al., Regardless of the level of exposure, no E. mathaei were recorded on 1993). However, this argument seems inadequate here as D. setosum the two MPAs, Chumbe and Mnemba. Predation is probably the preva- on Bawe present rates of bioerosion almost twice as high as those on lent regulatory agent affecting E. mathaei proliferation on marine- Changu, despite having similar average body size. Other currently unac- protected areas (McClanahan and Kurtis, 1991; McClanahan and counted for variables, such as the hardness or particular type of substra- Muthiga, 1989). Exclusion by predation can thus explain the sharp con- ta available for grazing, may account for these differences. trast between the extremely high E. mathaei densities observed on reefs Bioerosion rates on the exposed sites were highest on Pongwe, with −2 adjacent to the Mnemba MPA, in contrast to their absence from within more than 6.9 kg CaCO3 m eroded annually. The higher gross the protected zone. Alternatively, the effects of predation might still bioerosion on Pongwe compared to Kiwengwa, despite the higher total be evident even when echinoids are not completely excluded as in the sea urchin abundance on the latter site (ca. 15.5 and 20.3 ind m−2,re- case of Mnemba. Strong predation pressure of D. setosum in the MPA spectively), may be attributed to differences in species composition be- of Chumbe could be assumed from their bimodal size frequency distri- tween the two sites. While on Pongwe there are three species bution (Fig. S1), as mid-sized individuals are most susceptible to preda- comprising the reef-eroding echinoid guild, the reefs of Kiwengwa are tion while large predator-immune sea urchins and newly recruited solely occupied by E. mathaei. As different species may occupy different individuals are more likely to escape predation (Ojeda and Dearborn, niches on the reef, interspecific competition may be reduced allowing 1991; Scheibling and Hamm, 1991; Tegner and Levin, 1983). In unpro- more resources for feeding. Although intraspecific food consumption is tected reefs, the difference in E. mathaei proliferation between exposed expected to increase as sea urchins grow larger (Bak, 1990; Klumpp and sheltered reefs may also be attributed to human exploitation et al., 1993), the individual rates of food consumption for the larger through fishing. Reduced fishing success on sheltered reefs may be mean sized E. mathaei from Kiwengwa were lower than those from attributed to the relatively high coral cover and structural complexity Pongwe. The high E. mathaei abundance on Kiwengwa inevitably in- of these reefs, which may provide more refuge from fishing for potential creases intraspecific competition which consequently utilizes resources E. mathaei predators in comparison to bare exposed reefs (McClanahan, that might otherwise be channeled to feeding. 1997). Alternatively, the strong surf and currents associated with ex- The main differences in bioerosion between the eastern and western posed reefs may restrict fishing by forcing fishermen away from ex- reefs should rather be attributed to E. mathaei, which total erosion was posed sites or limit fishing duration, allowing more echinoid predators more than 12 times higher on eastern reefs. These rates are higher to avoid being caught. than those reported for E. mathaei from the Red Sea 0.5–0.9 kg CaCO3 m−2 year−2 (Mokady et al., 1996); Enewetak Atoll 0.08–0.33 kg −2 −2 4.3. Bioerosion and herbivory CaCO3 m year (Russo, 1980); Marshall Islands 3.3 kg CaCO3 m −2 −2 −2 −2 year (Russo, 1980); Kenya 1.3 kg CaCO3 m year on unprotect- Accurate assessments of the proportion of reworked sediments in ed reefs (Carreiro-Silva and McClanahan, 2001); and Moorea, French −2 −2 the diet of echinoids are still inconclusive, as evident from the variety Polynesia 0.37 kg CaCO3 m year (Bak, 1990), but lower than the of methodologies that have been used to obtain them. These methodol- exceptionally high rates reported from the Arabian Gulf 9.9–15.3 kg −2 −2 ogies, often yielding considerably different results, may include the use CaCO3 m year (Downing and El-Zahr, 1987). Another difference of petrographic sections of fecal pellets (Bak, 1990; Hunter, 1977; between exposed and sheltered reefs is the absence of D. setosum

Scofﬁn et al., 1980), CaCO3 evaluations in the guts of urchins from on the eastern reefs. However, the lack of D. setosum bioerosion is noncarbonated substrates (Mokady et al., 1996), and even comparisons diminished by the high bioerosion rates of E. mathaei. to other echinoids that are presumed to be non-eroding species One important contributing factor to different bioerosion rates is (Carreiro-Silva and McClanahan, 2001). Nonetheless, except for the body size. The larger body size of E. diadema may act in two ways to fa- biases that originate from using different methodologies, it is still largely cilitate its exceptionally high herbivory and CaCO3 consumption rates debated whether ingestion of loose sediment (reworked sediment) by (Table 2): (a) the larger Aristotle's lantern (the sea urchin's feeding ap- sea urchins is merely a by-product of their grazing activity (i.e. uninten- paratus) and intestines, associated with larger body size (Black et al., tional sediment ingestion) and as such a consequence of sediment loads, 1982; Ebert, 1980), may increase the volume of food ingestion and di- or an active strategy of intentional grazing that is governed by the life gestion; and (b) a larger body may reduce the risk of predation histories of the different species. In the current study, we have observed, (McClanahan and Muthiga, 1989), increasing the duration of food forag- on several occasions, groups of Diadema setosum actively feeding on ing at the expense of seeking shelter. In addition to quantity, body size 16 O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 may also determine the quality of the food consumed. For example, in Acknowledgments two similarly-sized Diadema populations (67.8 ± 6.2 mm and 69.9 ± 6.1 mm, mean ± SD, for D. setosum and D. savignyi, respectively) in This work was supported by the World Bank Group and the Global

Kenya, no difference was observed in the organic and CaCO3 portions Environmental Facility (GEF) through the Coral Reef Targeted Research of the two species (Carreiro-Silva and McClanahan, 2001). In contrast, and Capacity Building for Management program, Coral Bleaching asignificant size difference in similar species in Zanzibar (45.3 mm and Local Environmental Responses working group to YL. We thank and 53.8 ± 0.3 mm, mean ± SE, for D. setosum and D. savignyi,respec- the staff of the IMS in Tanzania for the use of their facilities. We are tively), revealed a higher proportion of organic matter in the larger grateful to Dr. A. Zvuloni for providing data on coral and macroalgae, D. savignyi. As larger Diadema individuals are potentially more mobile to Dr. N. Shenkar for her advice on an early draft of this manuscript, and less prone to predation, they may exploit feeding grounds inacces- and to N. Paz and M. Chen Bronstein for their editorial assistance. [ST] sible to smaller individuals, consequently maximizing organic matter consumption. The two MPAs, Chumbe on the west and Mnemba on the east, dif- References fered from neighboring unprotected sites in both echinoid species com- Andrew, N.L., 1993. Spatial heterogeneity, sea urchin grazing, and habitat structure on position and abundance, and consequently in the rates of bioerosion. reefs in temperate Australia. Ecology 74 (2), 292–302. For example, on Mnemba, echinoid densities were so low that sea- Bajkov, A.D., 1935. How to estimate the daily food consumption of fish under natural conditions. Trans. Am. Fish. Soc. 65 (1), 288–289. urchin-induced bioerosion was considered negligible. Similarly on Bak, R.P.M., 1990. Patterns of echinoid bioerosion in two Pacific coral reef lagoons. Mar. Chumbe, despite the presence of D. setosum, bioerosion rates were the Ecol. Prog. Ser. 66, 267–272. lowest recorded per site (excluding Mnemba). As predation controls Bak, R.P.M., 1994. Sea urchin bioerosion on coral reefs: place in the carbonate budget and – fi relevant variables. Coral Reefs 13 (2), 99 103. echinoid proliferation (McClanahan, 1998), and shing may in turn Bak, R.P.M., van Eys, G., 1975. Predation of the sea urchin Diadema antillarum Philippi on reduce potential predators (McClanahan et al., 1999), sea urchin popu- living coral. Oecologia 20 (2), 111–115. lations on highly exploited reefs are expected to proliferate. Conse- Bak, R.P.M., Carpay, M.J.E., de Ruyter van Steveninck, E.D., 1984. Densities of the sea urchin fi Diadema antillarum before and after mass mortalities on the coral reefs of Curaçao. quently, except for protected areas, where shing restrictions are Mar. Ecol. Prog. Ser. 17, 105–108. enforced, extensive reef exploitation may have favored sea urchin pro- Bauer, J.C., 1980. Observations on geographical variations in population density of the liferation, which is accelerating reef degradation through bioerosion. echinoid Diadema antillarum within the western North Atlantic. Bull. Mar. Sci. 30 – In addition to regulation of community structure, the type of sub- (2), 509 515. Bellwood, D.R., Hughes, T.P., Folke, C., Nyström, M., 2004. Confronting the coral reef crisis. stratum has also been shown to directly affect echinoid rates of Nature 429 (6994), 827–833. bioerosion. Hibino and Van Woesik (2000), for example, found sub- Bergman, K.C., Öhman, M.C., 2001. Coral reef structure at Zanzibar Island, Tanzania. In: strate age to be an important factor affecting bioerosion rates by Richmond, M.D., Francis, J. (Eds.), Proc. 20th Anniversary Conference on Advances in Marine Science in Tanzania. Marine Science Development in Tanzania and Eastern E. mathaei: older substrates lost on average more carbonate than did re- Africa, pp. 263–275. cently dead coral on experimental tiles. As echinoid populations on Birkeland, C., Randall, R.H., 1981. Facilitation of coral recruitment by echinoid excavations. sheltered reefs, in the current study, inhabit fairly flourishing reefs, Proceedings of the Fourth International Coral Reef Symposium, Manila, pp. 695–698. Black, R., Johnson, M.S., Trendall, J.T., 1982. Relative size of Aristotle's lantern in they encounter mostly young substrata (mostly coral and coralline Echinometra mathaei occurring at different densities. Mar. Biol. 71 (1), 101–106. algae) of recently accreted origin. In contrast, the exposed reefs on Briscoe, C.S., Sebens, K.P., 1988. Omnivory in Strongylocentrotus droebachiensis (Müller) Zanzibar's eastern side are showing continued signs of deterioration (Echinodermata: Echinoidea): predation on subtidal mussels. J. Exp. Mar. Biol. Ecol. 115 (1), 1–24. and loss of hard-coral cover (Mbije et al., 2002). Thus, with little accre- Carpenter, R.C., 1981. Grazing by Diadema antillarum (Philippi) and its effects on the tion to compensate for the loss of substrata, echinoid bioerosion is ex- benthic algal community [sea urchin damage]. J. Mar. Res. 39, 749–765. pected to accelerate with time, as older and faster eroding parts of the Carpenter, R.C., Edmunds, P.J., 2006. Local and regional scale recovery of Diadema pro- – reef are constantly being exposed. Apart from sea urchins already motes recruitment of scleractinian corals. Ecol. Lett. 9 (3), 271 280. Carreiro-Silva, M., McClanahan, T.R., 2001. Echinoid bioerosion and herbivory on Kenyan being key drivers of the carbonate budget on most coral reefs (Scoffin coral reefs: the role of protection from fishing. J. Exp. Mar. Biol. Ecol. 262 (2), et al., 1980), as coral cover declines the system will become less sensi- 133–153. fl tive to drivers of calcification and ever more sensitive to echinoid Clemente, S., Hernández, J.C., 2008. In uence of wave exposure and habitat complexity in determining spatial variation of the sea urchin Diadema aff. antillarum (Echinoidea: bioerosion (Perry et al., 2012). Moreover, intense erosion of CCA will Diadematidae) populations and macroalgal cover (Canary Islands—Eastern Atlantic not just eliminate a significant ingredient of the reef framework, but Ocean). Rev. Biol. Trop. 56 (3), 229–254. may also hinder reef recovery as some species are known to chemically Dean Jr., W.E., 1974. Determination of carbonate and organic matter in calcareous sediments and sedimentary rocks by loss on ignition: comparison with other methods. induce recruitment of corals and octocorals (Harrington et al., 2004; J. Sediment. Res. 44 (1), 242–248. Heyward and Negri, 1999; Morse et al., 1996). Downing, N., El-Zahr, C.R., 1987. Gut evacuation and filling rates in the rock-boring sea Sea urchin erosion may temporarily act to increase the complexity of urchin, Echinometra mathaei.Bull.Mar.Sci.41,579–584. Dumas, P., Kulbicki, M., Chifflet, S., Fichez, R., Ferraris, J., 2007. Environmental factors exposed reefs by enhancing to the 3-dimensional structure of abraded influencing urchin spatial distributions on disturbed coral reefs (New Caledonia, benthic environments, consequently facilitating increased diversity in South Pacific). J. Exp. Mar. Biol. Ecol. 344 (1), 88–100. these habitats. However, as echinoid populations continue to prolifer- Ebert, T.A., 1980. Relative growth of sea urchin jaws: an example of plastic resource allo- cation. Bull. Mar. Sci. 30 (2), 467–474. ate, resources will eventually dwindle, and intensifying competition Ebert, T.A., 1982. Longevity, life history, and relative body wall size in sea urchins. Ecol. might work to exclude other benthic species, leading to reduced biodi- Monogr. 52 (4), 353–394. versity. Consequently, in areas of low coral cover and little regulation of Elliott, J.M., 1972. Rates of gastric evacuation in brown trout, Salmo trutta L. Freshw. Biol. 2 (1), 1–18. echinoid proliferation, such as on Zanzibar's exposed unprotected reefs, Elliott, J.M., Persson, L., 1978. The estimation of daily rates of food consumption for fish. J. sea urchin bioerosion may ultimately lead to a negative carbonate bud- Anim. Ecol. 47 (3), 977–991. get and consequently to reef degradation. Our data suggest that these Glynn, P.W., Wellington, G.M., Birkeland, C., 1979. Coral reef growth in the Galapagos: – processes are happening faster on exposed rather than sheltered reefs. limitation by sea urchins. Science 203 (4375), 47 49. Graham, N.A.J., Nash, K.L., 2013. The importance of structural complexity in coral reef eco- These differences may be attributed to the complete dominance of the systems. Coral Reefs 32 (2), 315–326. sea urchin E. mathaei, having higher densities and a larger average Griffin, S.P., García, R.P., Weil, E., 2003. Bioerosion in coral reef communities in southwest – body size than on sheltered reefs. Nonetheless, the establishment of ma- Puerto Rico by the sea urchin Echinometra viridis. Mar. Biol. 143 (1), 79 84. Harrington, L., Fabricius, K., De'ath, G., Negri, A., 2004. Recognition and selection of settle- rine protected areas can dramatically slow this process, and facilitate ment substrata determine post-settlement survival in corals. Ecology 85 (12), the sustainability of the reef framework by providing efficient regula- 3428–3437. tion over echinoid populations. Hawkins, C.M., Lewis, J.B., 1982. Ecological energetics of the tropical sea urchin Diadema antillarum Philippi in Barbados, West Indies. Estuar. Coast. Shelf Sci. 15 (6), 645–669. Supplementary data to this article can be found online at http://dx. Heyward, A.J., Negri, A.P., 1999. Natural inducers for coral larval metamorphosis. Coral doi.org/10.1016/j.jembe.2014.03.003. Reefs 18 (3), 273–279. O. Bronstein, Y. Loya / Journal of Experimental Marine Biology and Ecology 456 (2014) 8–17 17

Hibino, K., van Woesik, R., 2000. Spatial differences and seasonal changes of net carbonate Ngoile, M.A.K., 1990. Ecological baseline surveys of coral reefs and intertidal zones around accumulation on some coral reefs of the Ryukyu Islands, Japan. J. Exp. Mar. Biol. Ecol. Mnemba Island and Zanzibar Town. Zanzibar Environ. Study Ser, 9. Commission for 252, 1–14. Lands and the Environment, Zanzibar. Hughes, T.P., 2003. Climate change, human impacts, and the resilience of coral reefs. Ngoile, M.A.K., Horrill, C.J., 1993. Coastal ecosystems, productivity and ecosystems protec- Science 301 (5635), 929–933. tion: coastal ecosystem management. Ambio 22, 461–467. Hunter, I.G., 1977. Sediment production by Diadema antillarum on a Barbados fringing Ngoile, M.A.K., Bwathondi, P.O.J., Makwaia, E.S., 1988. Trends in the exploitation of marine reef. Proceedings, Third International Coral Reef Symposium. University of Miami, fisheries resources in Tanzania. In: Mainoya, J.R. (Ed.), Ecology and Bioproductivity of Miami, Florida, pp. 106–109. the Marine and Coastal Waters of East Africa. University of Dar es Salaam, Dar es Jiddawi, N., 1997. The reef dependent fisheries of Zanzibar. In: Johnstone, R.W., Francis, J., Salaam, pp. 93–100. Muhando, C.A. (Eds.), National Conference on Coral Reefs, Zanzibar, Tanzania, Nordlund, L., Walther, A., 2010. Chumbe Island Coral Park — Conservation and Education pp. 22–35. Programme, Zanzibar. 1–36. Jiddawi, N.S., Öhman, M.C., 2002. Marine fisheries in Tanzania. AMBIO: J. Hum. Environ. Obura, D., Celliers, L., Machano, H., Mangubhai, S., Mohammed, M.S., Motta, H., Muhando, 31 (7), 518–527. C., Muthiga, N., Pereira, M., Schleyer, M., 2002. Status of Coral Reefs in Eastern Africa: Johansson, C.L., Bellwood, D.R., Depczynski, M., Hoey, A.S., 2013. The distribution of the Kenya, Tanzania, Mozambique and South Africa. 63–78. sea urchin Echinometra mathaei (de Blainville) and its predators on Ningaloo Reef, Ojeda, P.F., Dearborn, J.H., 1991. Feeding ecology of benthic mobile predators: experimen- Western Australia: the implications for top–down control in an intact reef system. tal analyses of their influence in rocky subtidal communities of the Gulf of Maine. J. J. Exp. Mar. Biol. Ecol. 442, 39–46. Exp. Mar. Biol. Ecol. 149 (1), 13–44. Johnson, M.P., Frost, N.J., Mosley, M.W.J., Roberts, M.F., Hawkins, S.J., 2003. The area- O'leary, J.K., McClanahan, T.R., 2010. Trophic cascades result in large-scale coralline algae independent effects of habitat complexity on biodiversity vary between regions. loss through differential grazer effects. Ecology 91 (12), 3584–3597. Ecol. Lett. 6 (2), 126–132. Perry, C.T., Edinger, E.N., Kench, P.S., Murphy, G.N., Smithers, S.G., Steneck, R.S., Mumby, P. Khatib, A.H., 1997. The importance of tourism on coral reefs in Zanzibar. In: Johnstone, R. J., 2012. Estimating rates of biologically driven coral reef framework production and W., Francis, J., Muhando, C.A. (Eds.), National Conference on Coral Reefs, Zanzibar, erosion: a new census-based carbonate budget methodology and applications to Tanzania, pp. 35–37. the reefs of Bonaire. Coral Reefs 31 (3), 853–868. Klinger, T.S., Lawrence, J.M., Lawrence, A.L., 1997. Gonad and somatic production of Peyrot-Clausade, M., Chabanet, P., Conand, C., Fontaine, M.F., Letourneur, Y., Harmelin- Strongylocentrotus droebachiensis fed manufactured feeds. Bull. Aquac. Assoc. Can. Vivien, M., 2000. Sea urchin and fish bioerosion on La Reunion and Moorea reefs. 97 (1), 35–37. Bull. Mar. Sci. 66 (2), 477–485. Klumpp, D.W., Salita-Espinosa, J.T., Fortes, M.D., 1993. Feeding ecology and trophic role of Richmond, M.D., 2002. A Field Guide to the Seashores of Eastern Africa and the Western sea urchins in a tropical seagrass community. Aquat. Bot. 45 (2–3), 205–229. Indian Ocean Islands, 2nd ed. Sida/SAREC - UDSM (461 pp.). Langdon, M.W., 2012. The Ecology of the Grazing Urchin Echinometra mathaei at Ningaloo Russo, A.R., 1977. Water flow and the distribution and abundance of echinoids (genus Marine Park. Faculty of Science and Engineering. Murdoch University, Perth, Australia Echinometra) on an Hawaiian reef. Mar. Freshw. Res. 28 (6), 693–702. 382. Russo, A.R., 1980. Bioerosion by two rock boring echinoids (Echinometra mathaei and Larson, B.R., Vadas, R.L., Keser, M., 1980. Feeding and nutritional ecology of the sea urchin Echinostrephus aciculatus) on Enewetak Atoll, Marshall Islands. J. Mar. Res. 38 (1), Strongylocentrotus drobachiensis in Maine, USA. Mar. Biol. 59 (1), 49–62. 99–110. Lawrence, J.M., 1975. On the relationship between marine plants and sea urchins. Sammarco, P.W., 1980. Diadema and its relationship to coral spat mortality: grazing, com- Oceanogr. Mar. Biol. Annu. Rev. 13, 213–286. petition, and biological disturbance. J. Exp. Mar. Biol. Ecol. 45 (2), 245–272. Lawrence, J.M., Sammarco, P.W., 1982. Effect of feeding on the environment: Echinoidea. Sammarco, P.W., 1982. Echinoid grazing as a structuring force in coral communities: In: Jangoux, M., Lawrence, J.M. (Eds.), Echinoderm Nutrition. A. A. Balkema Press, whole reef manipulations. J. Exp. Mar. Biol. Ecol. 61 (1), 31–55. Rotterdam, pp. 499–519. Scheibling, R.E., Hamm, J., 1991. Interactions between sea urchins (Strongylocentrotus Mapstone, B.D., Andrew, N.L., Chancerelle, Y., Salvat, B., 2007. Mediating effects of sea droebachiensis) and their predators in field and laboratory experiments. Mar. Biol. urchins on interactions among corals, algae and herbivorous fish in the Moorea 110 (1), 105–116. lagoon, French Polynesia. Mar. Ecol. Prog. Ser. 332, 143–153. Scoffin, T.P., Stearn, C.W., Boucher, D., Frydl, P., Hawkins, C.M., Hunter, I.G., MacGeachy, J.K., Mbije, N.E., Wagner, D., Francis, J., Öhman, M.C., Garpe, K., 2002. Patterns in the distribu- 1980. Calcium carbonate budget of a fringing reef on the west coast of Barbados. tion and abundance of hard corals around Zanzibar Island. Ambio 31 (7–8), 609–611. Part II — erosion, sediments and internal structure. Bull. Mar. Sci. 30 (2), 475–508. McClanahan, T.R., 1988a. Coexistence in a sea urchin guild and its implications to coral Siddon, C.E., Witman, J.D., 2003. Influence of chronic, low-level hydrodynamic forces on reef diversity and degradation. Oecologia 77 (2), 210–218. subtidal community structure. Mar. Ecol. Prog. Ser. 261, 99–110. McClanahan, T.R., 1988b. Seasonality in East Africa's coastal waters. Mar. Ecol. Prog. Ser. Steneck, R.S., 2013. Sea urchins as drivers of shallow benthic marine community struc- 44, 191–199. ture, In: Lawrence, J.M. (Ed.), Sea Urchins: Biology and Ecology, 3rd ed. Academic McClanahan, T., 1997. Effects of fishing and reef structure on East African coral reefs. Proc Press, San Diego. 8th Int Coral Reef Sym, pp. 1533–1538. Team, R.D.C., 2013. R: A Language and Environment for Statistical Computing. R Founda- McClanahan, T.R., 1998. Predation and the distribution and abundance of tropical sea tion for Statistical Computing, Vienna, Austria (Retrieved from http://www.R-project. urchin populations. J. Exp. Mar. Biol. Ecol. 221 (2), 231–255. org). McClanahan, T.R., Kurtis, J.D., 1991. Population regulation of the rock-boring sea urchin Tegner, M.J., Levin, L.A., 1983. Spiny lobsters and sea urchins: analysis of a predator–prey Echinometra mathaei (de Blainville). J. Exp. Mar. Biol. Ecol. 147 (1), 121–146. interaction. J. Exp. Mar. Biol. Ecol. 73 (2), 125–150. McClanahan, T.R., Muthiga, N.A., 1989. Patterns of predation on a sea urchin, Echinometra Trudgill, S.T., Smart, P.L., Friederich, H., Crabtree, R.W., 1987. Bioerosion of intertidal lime- mathaei (de Blainville), on Kenyan coral reefs. J. Exp. Mar. Biol. Ecol. 126 (1), 77–94. stone, Co. Clare, Eire — 1: Paracentrotus lividus. Mar. Geol. 74 (1–2), 85–98. McClanahan, T.R., Muthiga, N.A., 2001. The ecology of Echinometra.In:Lawrence,J.M.(Ed.), Tuya, F., Cisneros-Aguirre, J., Ortega-Borges, L., Haroun, R.J., 2007. Bathymetric segregation Developments in Aquaculture and Fisheries Science. Elsevier, pp. 225–243. of sea urchins on reefs of the Canarian Archipelago: role of flow-induced forces. McClanahan, T.R., Shafir, S.H., 1990. Causes and consequences of sea urchin abundance Estuar. Coast. Shelf Sci. 73 (3–4), 481–488. and diversity in Kenyan coral reef lagoons. Oecologia 83 (3), 362–370. Vadas, R.L., Elner, R.W., 2003. Responses to predation cues and food in two species of sym- McClanahan, T.R., Muthiga, N.A., Kamukuru, A.T., Machano, H., Kiambo, R.W., 1999. The patric, tropical sea urchins. Mar. Ecol. 24 (2), 101–121. effects of marine parks and fishing on coral reefs of northern Tanzania. Biol. Conserv. Vaïtilingon, D., Rasolofonirina, R., Jangoux, M., 2003. Feeding preferences, seasonal gut re- 89 (2), 161–182. pletion indices, and diel feeding patterns of the sea urchin Tripneustes gratilla McClintock, J.B., Klinger, T.S., Lawrence, J.M., 1982. Feeding preferences of echinoids for (Echinodermata: Echinoidea) on a coastal habitat off Toliara (Madagascar). Mar. plant and animal food models. Bull. Mar. Sci. 32 (1), 365–369. Biol. 143, 451–458. McCook, L., Jompa, J., Diaz-Pulido, G., 2001. Competition between corals and algae on Wheeler, B., 2010. lmPerm: permutation tests for linear models. http://CRAN.R-project. coral reefs: a review of evidence and mechanisms. Coral Reefs 19 (4), 400–417. org/package=lmPerm. Mills, S.C., Peyrot-Clausade, M., Fontaine, M.F., 2000. Ingestion and transformation of algal Young, M.A.L., Bellwood, D.R., 2011. Diel patterns in sea urchin activity and predation on turf by Echinometra mathaei on Tiahura fringing reef (French Polynesia). J. Exp. Mar. sea urchins on the Great Barrier Reef. Coral Reefs 30 (3), 729–736. Biol. Ecol. 254 (1), 71–84. Zanzibar Department of Environment MACEMP, 2009. In: Department of Environment Mokady, O., Lazar, B., Loya, Y., 1996. Echinoid bioerosion as a major structuring force of (Ed.), The status of Zanzibar costal resources. Towards the development of integrated Red Sea coral reefs. Biol. Bull. 190 (3), 367–372. coastal management strategies and action plan, p. 104 (Dar es Salaam). Morse, A.N.C., Iwao, K., Baba, M., Shimoike, K., Hayashibara, T., Omori, M., 1996. An ancient chemosensory mechanism brings new life to coral reefs. Biol. Bull. 191 (2), 149–154.